Questions we would like to learn (scattered through the whole lecture)

Climate Impacts on the Baltic Sea: From Science to Policy Bornholm, July 2009 (Paleo) Climate Modelling Eduardo Zorita GK SS Research Centre, Geesthacht, Germany

Questions we would like to learn (scattered through the whole lecture) 1. What are a climate models (for)? Simulate and understand the past, present and future climate Make experiments with a virtual Earth 2. What climate models can and cannot do Simulates some aspects very well Others not so well 3. Climate variations of the Late Holocene and their causes Holocene cooling? The Medieval Warm Period ~ 1200 AD The Little Ice Age ~ 1700 AD Recent warming

Average simulated surface temperature response 3.8 K 1.4 K

Climate sensitivity and climate feedbacks Climate sensitivity: Change in equilibrium global temperature after an initial external forcing has changed by a certain amount Climate feedback: to what extent the change in global temperature modifies the external forcing Climate feedbacks determine climate sensitivity

Main climate forcings in the period 7000BP to present are: -Orbital forcing, due to slow changes in the Earth's orbital parameters. The precession of the perihelion (period ca. 19000 years), obliquity (period ca. 40000 years) and eccentricity (period ca. 100000 years). This forcing can be accurately calculated. -CO2 and CH4 concentrations. Derived from the concentrations in air bubbles trapped in ice cores -Intrinsic solar irradiance, caused by internal solar dynamics. Derived from concentrations of the isotopes C14 and Be10 in ice cores. Produced by cosmic rays, their production rate is modulated by the solar open magnetic field -Volcanic forcing, caused by the production of stratospheric aerosols from sulphate volcanic eruptions

Rotation Precession Nutation in obliquity http://en.wikipedia.org/wiki/nutation

Solar Insolation Insolation is the amount of solar energy that strikes the top of the atmosphere per unit area and per unit of time. It varies with latitude and season. solation were calculated by A. Berger /Belgium. Data includes both the 1978 calculations and the latest (1991) solution published in Quaternary Science Reviews.

Holocene insolation changes, as a function of season, latitude and time Wanner et al., The Holocene (2008)

External forcings Derived from C14 concentrations in tree rings, Weber and Crowley (2004), and from Be10 in ice cores, Crowley (2000).

External forcings Derived from C14 concentrations in tree rings, Weber and Crowley (2004) and from Be10 in ice cores, Crowley (2000).

0.30 % Ice core acidity 10Be + sun spots

What is the greenhouse gas forcing?

~ 36 328 Instantaneous doubling of CO2 concentrations, everything else fixed

The Earth has multiple ways to react, and to simulate this reaction is difficult -Increase surface temperature -Increase evaporation -Increase cloudiness -Increase ocean heat-storage -All of them simultaneously

Due to the near saturation of absorption bands the greenhouse forcing is proportional to the logarithm of CO2 concentrations

Average simulated surface temperature response 3.8 K 1.4 K

Climate models replicate the observed global T evolution using observed forcings -Uncertainty in aerolsol forcing -Different climate model sensitivity

External forcings of 20th century climate change (without volcanoes) }? Total anthropogenic forcing in 2100 : 6.7 (4.2-9.1) w/m2

Each point represents a climate model Forcing arbitrary chosen { within uncertainty bounds There should be no relationship Kiehl, GRL 2007 Intrinsic property of each model

High latitude near surface temperature in the last 70 mill. years J. Zachos et al., Science 292, 686-693 (2001) M Pagani et al. Nature 460, 85-88 (2009) doi:10.1038/nature08133

High latitude temperature+ global ice volume in the last 35 mill. years Zachos et al., Science 292, 686-693 (2001)

Pleistocene Climate variations recorded in Antarctic Ice also identified in marine sediment-cores 280 ppm global T ~ 6 K 180 ppm 120 meters { {

How did the Earth surface look like at the Glacial Maximum? Global temperature difference ~ 7 K CO2 atmospheric concentration difference ~ 100 ppm

Greenland ice core records showing sub Milankovic variability: the Dansgaard Oeschger events and Heinrich events

North Reconstructed European Holocene temperatures from pollen assemblages East West Northern and Central Europe warmer in the Mid-Holocene Optimum Winter Summer South Davis et al. Quat. Sci. Rev 2003

Reconstructed Northern Hemisphere temperatures in the past two millennia Medieval Warm Period ->Little Ice Age ->Recent Warming

Recent global warming ~ 0.7 Kelvin Global land warming ~ 1 Kelvin

Recent global warming ~ 0.7 Kelvin Global land warming ~ 1 Kelvin GHG natural variability+ GHG+solar Tropospheric aerosols +GHG+ solar

Methods to estimate global temperatures Long distance past (106 105 years): concentrations of O18 in plankton skeletons identified in sediment cores. Resolution ~ 103-10 4, limited by deposition rate. Dating by estimation of depositions rate and geological markers. Recent-distance past ( 105-104 years) : O18 concentration in ice cores. Resolution ~102-103, limited by precipitation rate. Dating by layer counting and ice-flow model. Recent past (103 year): pollen assemblages in lake or coastal sediments. Resolution ~ 102. Limited by deposition rates. Dating by C14 Historical past (102-103 years): dendro-climatological data. Resolution annual, exact by ring counting Instrumental period: Stations readings, satellite remote sensing

The climate as a heat machine

The complexity of the climate system

Some important climate feedbacks (+,-,unknown, in theory) Black-body: increased temperatures increase the outward long-wave emission Water vapor feedback: Warmer ocean temperatures increase evaporation -> atmospheric humidity -> water vapor greenhouse forcing Cloud feedback: warmer temperatures change cloud cover -> short wave and long wave radiation forcing. Sign depends on cloud type, cloud location Surface albedo: warmer temperatures melt snow and ice -> albedo decrease Lapse-rate feedback: decreased vertical temperature profile decrease the atmospheric greenhouse gas forcing Many other feedbacks involve vegetation, soil moisture, oceanic circulation, carbon cycle, etc

Structure of a General Circulation Model

Structure of a General Circulation Model courtesy D. Vinner, CRU

What is (in) a climate model? A computer program (0.5 mill pages) that was written to represent: air flows from high pressure to low pressure the Earth is round and rotates hot air is lighter than cold air solar radiation is absorbed and reflected by all materials infrared radiation is absorbed and emitted by all materials water vapor condenses below certain relative humidity threshold, clouds are formed. And it may rain Warm water warms the air, warm air warms the water surface Rain makes sea water fresher, evaporation makes seawater saltier water masses flow from high pressure to low pressure winds exert a drag on ocean surface. currents arise warm water is lighter than cold water salt water is heavier than fresh water

Example of output of a climate model Complete output is similar to that of a weather prediction model: 3 d T, 3 d wind, 3 d moisture, 3 d cloudiness, precipitation, A very complete,global, huge data set

Some numbers of climate models - Spectral (spherical harmonics) or finite differences schemes -Fortran code with some pieces of C code- ~ 5x105 lines -On non-massive parallel machines (101 nodes), resolution of 2.5x2.5 degrees -Size of model output -Typical length of simulations: 1 model -year takes ~ 4 hours ~ 1 GB per model-year 100-1000 years

Dynamical atmospheric processes

Dynamical atmospheric processes represented in a global climate model Parametrization: heuristic, semiempirical representation: For instance: if a grid cell is 80% saturated and air is rising it will be 60% covered by clouds, it will rain 2.3 mm/day But no individual cloudsl are simulated The set of all semi empirical parameters is called model parametrization. It modulates the sensitivity of the model to external perturbations

Dynamical oceanic processes

Dynamical oceanic processes represented in a global climate model

About 65% of the Earth surface is covered by clouds at any time Tropic s Extratropics

Average simulated surface temperature response 3.8 K 1.4 K

Regional climate predictions: considering the uncertainties

Simulation of the Iberian climate in the past 500 years with ECHO G >> MM5 Project Ramshes, Spain Simulation of the Iberian Climate in the past 500 years One way double nesting embedding : ECHO G >> Domain 1 >> Domain 2 ECHO G (~ 3.75x3.75 degrees) 145x145 km 145x145 km 45x45k m 45x45k m

Iberian winter precipitation, simulated, observed and reconstructed

Global climate simulation of the past millennium Model ECHO-G ERIK-1 simulation Erik the red, around 950 A.D ECHO-G Model used in IPCC AR4 Atmospheric model ECHAM4 19 levels, 3.75x3.75 deg res. Ocean model HOPE 20 levels, 2 x2 deg res. Flux correction applied spatial average zero

Plausible millennial evolution of Iberian temperature and winter precipitation temperature precipitation

Plausible millennial evolution of Iberian temperature and winter precipitation temperature Medieval Warm Period precipitation

Plausible millennial evolution of Iberian temperature and winter precipitation precipitation temperature Little Ice Age

Plausible millennial evolution of Iberian temperature and winter precipitation temperature precipitation Recent Warm Period

Plausible millennial evolution of Iberian temperature and winter precipitation precipitation temperature Warmer Drier Colder More humid Warmer Drier

Plausible millennial evolution of Iberian temperature and winter precipitation precipitation temperature Warmer Drier Colder More humid Did this really happen in the past? Warmer Drier

Reconstruction of Stockholm winter-spring temperature from tax records in the past 500 years

Sources of uncertainties... (very important) External forcing (solar, volcanic, land use) is uncertain Model (different models may yield different answers) Even using one single, different simulations may yield different answers: internal variability

External and internal variability Different initial conditions imply a different path along the whole simulation

External and internal variability Different initial conditions imply a different path along the whole simulation The path taken by a particular simulation depends: on the external forcing > deterministic component (external variability) on the initial conditions chosen. The real initial conditions, e.g. in year 800 A.D. are not known > stochastic component (natural variability)

External and internal variabiliry Different initial conditions imply a different path along the whole simulation The path taken by a particular simulation depends: on the external forcing > deterministic component (external variability) on the initial conditions chosen. The real initial conditions, e.g. in year 800 A.D. are not known > stochastic component (natural variability) Therefore, we cannot compare directly observed and simulated paths This comparison can be only successful at timescales where the variations are caused by the external forcing. Which are these timescales? Not exactly known. Hopefully, multidecadal...

Holocene simulations

ECHAM4 HOPE G MODEL ECHAM4 spectral T30 (global,aprox. 3.75 x 3.75 deg. resolution), 19 levels highest level 10 hpa, 4 5 levels in the stratosphere HOPE: primitive equation ocean ECHO G land mask for the Atlantic European sector model, equivalent resolution 2.8x 2.8 with equatorial grid refinement, 20 levels Coupled through OASIS, flux adjustment to avoid long term drift constant in time and zero spatial average Model timestep: 30 minutes, data stored each 12 model hours

Global climate simulations Historical forcing Scenario B2 Erik 4 Scenario A2 Erik 1 Erik 2 1000 A.D. 1756 A.D. 1990 A.D. Erik 3 2100 A.D.

Global climate simulations 8.2K Event 7000BP 6000 BP 4000BP 1000 BP Holocene optimum present LMM DMM 3.75 *3.75 GCM atmospheric part oceanic part ECHAM 4 -HOPE G ECHO-G horizontal resolution (~200km longitude, ~400km latitude) Oetzi 1 variations in orbital forcing (derived from astronomical calculations) (solar irradiance and greenhouse gases kept constant) Oetzi 2 variations in orbital forcing, solar irradiance and greenhouse gases (derived from ice core information) Erik 1-4

Simulated Holocene global mean temperature

latitude north (degrees) ECHO-G model Baltic Area atmospheric grid longitude east (degrees)

ECHO-G (Oetzi-II) Baltic Area simulated near-surface air temperature and precipitation

ECHO-G (Oetzi-II) simulated NAO-Index

Around 1000 A.D? Scandinavian summer temperature reconstruction Grudd, Clim.Dyn (2008)